AMD unleashes first ever commercial “5GHz” CPU, the FX-9590

[toyota]

I don't know why people are so upset about breaking the obsolete 95/130W TDP on a desktop enthusiast CPU. My GPU draws 265W and people are calling it an efficient and a cool running card, so give me a break. I, for one, would welcome 225W TDP on CPUs just like we've had on GPU for years. It's not 2004, we have cooling solution that can deal with that much heat. My card has no problems with dissipating 265W of power and it's way easier to cool a CPU then a GPU. I'd welcome a stock 225W TPD 8-core Sandy-E running at 5GHz.

please explain that one...

 

[guskline]

Thanks for that. I have a request though. Would you mind running Povray 3.7 benchmark with 1 thread and 8 threads and reporting those results? I have a 1045T which I'll test with too (1 thread, 6 threads).

Per your request I ran POV Ray ver 3.7 RC7 on the 8350 stock and at 4.7 Ghz and the 3770k stock and at my OC of 4.4Ghz. I ran a 1 thread render and a 8 thread render. Here we go with results.
For 1 thread
8350 stock 15 minutes 50seconds
8350 @ 4.7 Ghz 15m 47s
3770k stock 15m12s
3770k @ 4.4Ghz 12m 29s

For 8 threads
8350 stock 2m30s
8350 @4.7Ghz 2m31s
3770k stock 3m17s
3770k @4.4Ghz 2m 43s

Hope this was the info you were looking for.

 

[frozentundra123456]

Well my claim that the FX-9590 will be faster than the 3930k and the 3970X was based in the average FPS. If we look at the mins, the 8350 gives 47 FPS and the 9590 would give about 54 FPS which places it between 3930k and the 3970X.



No. I was merely noticing how the measured 1.5% gain fits in the 1-5% prediction. Only those waiting 15% gain will be disappointed with my "logic".

Take a look to the FPS for the FX-4320 and the FX-8350. Both at 4.0GHz. Now ask yourself why doubling the cores increases the FPS by a 1.6% and you will get half the answer to why your point is invalid.

If you are making predictions on the basis of that graph, you cant predict cpu scaling from a gpu limited benchmark in any case. Granted there is a small difference among the cpus, but the test is also running into a gpu bottleneck, so any prediction of cpu scaling would be a wild guess at best.

 

[frozentundra123456]

Per your request I ran POV Ray ver 3.7 RC7 on the 8350 stock and at 4.7 Ghz and the 3770k stock and at my OC of 4.4Ghz. I ran a 1 thread render and a 8 thread render. Here we go with results.
For 1 thread
8350 stock 15 minutes 50seconds
8350 @ 4.7 Ghz 15m 47s
3770k stock 15m12s
3770k @ 4.4Ghz 12m 29s

For 8 threads
8350 stock 2m30s
8350 @4.7Ghz 2m31s
3770k stock 3m17s
3770k @4.4Ghz 2m 43s

Hope this was the info you were looking for.

Interesting. The 8350 doesnt seem to respond to overclocking at all in this test, while the 3770k responds pretty much as one would expect.

 

[guskline]

Interesting. The 8350 doesnt seem to respond to overclocking at all in this test, while the 3770k responds pretty much as one would expect.

I noticed the same thing. Weird. I hope I ran it correctly. Either way the 8 integer cores of the PileDriver flex their muscle in this test BTW any other benchmarks you want me to run for the 8350 vs the 3770k? Does anyone have a 4770k that they can run stock and at @4.3Ghz with the benchies I've run so far (Aida64, Passmark8 cpu only, cinebench 11.5 and POV rat 3.7)?

 

[notty22]

I recall others having benchmarking irregularities when o/c the 8350 also. It was throttling, and not from heat it seems. IMO, that's what will probably keep those m/b's alive that get the new 220 TDP chips.

 

[monstercameron]

Per your request I ran POV Ray ver 3.7 RC7 on the 8350 stock and at 4.7 Ghz and the 3770k stock and at my OC of 4.4Ghz. I ran a 1 thread render and a 8 thread render. Here we go with results.
For 1 thread
8350 stock 15 minutes 50seconds
8350 @ 4.7 Ghz 15m 47s
3770k stock 15m12s
3770k @ 4.4Ghz 12m 29s

For 8 threads
8350 stock 2m30s
8350 @4.7Ghz 2m31s
3770k stock 3m17s
3770k @4.4Ghz 2m 43s

Hope this was the info you were looking for.

so this is the performance difference that intel users on the forums have been denouncing the amd fx cpus for...38 sec difference?

oh the power draw is double you say...well how much power does your titan draw? or your crossfire-x draw? how long is it being fully loaded to make a difference while not idling?

Call me a "fanboi" but why would I buy the equivalent, more expensive Intel part when AMD parts are nearly as fast for a considerable amount less?

I mean a 3770k costs greater than $300 while a fx 8350 costs under $200...or am I missing something.

Even if the TDP is 220w what does that mean for you[reading this comment]? does how often do you run applications that fully load all cores for a considerable amount of time?

 

[FwFred]

Not my cup of tea, but it'll be interesting to see if there is any demand for something like this in the market. I guess that'll somewhat depend on system cost, which may be somewhat premium given the power and cooling requirements.

 

[Lepton87]

please explain that one...

I use my rig as an example:
CPU: cooler

Weight:1070 g (1240 g with 2 fans) Effectively 3(in this case even four) fans because it looks like that in a case

Titan's cooler. Do I need to explain further?

 

[frozentundra123456]

so this is the performance difference that intel users on the forums have been denouncing the amd fx cpus for...38 sec difference?

oh the power draw is double you say...well how much power does your titan draw? or your crossfire-x draw? how long is it being fully loaded to make a difference while not idling?

Call me a "fanboi" but why would I buy the equivalent, more expensive Intel part when AMD parts are nearly as fast for a considerable amount less?

I mean a 3770k costs greater than $300 while a fx 8350 costs under $200...or am I missing something.

Even if the TDP is 220w what does that mean for you[reading this comment]? does how often do you run applications that fully load all cores for a considerable amount of time?

Well, "if the shoe fits" as they say. It is only one benchmark though, and one of those in which the FX shows most favorably.

 

[toyota]

I use my rig as an example:
CPU: cooler
https://encrypted-tbn2.gstatic.com/...WtFsweZGFEjT_Tpl60-KAkQidZguAtOJUfR8ezOH-5a3w

Weight:1070 g (1240 g with 2 fans) Effectively 3(in this case even four) fans because it looks like that in a case
http://www.pctunerup.com/up/results/_201109/20110917204511_NH-D14shot2.jpg

http://images.anandtech.com/doci/6760/GeForceGTX_Titan_3Qtr2a_575px.jpg

Titan's cooler. Do I need to explain further?

yes you do because that gigantic behemoth cpu cooler is only designed to cool current oced cpus. so just what in the hell do you think it will take to cool on oced cpu that is already 220 watts out of the box?

 

[Lepton87]

yes you do because that gigantic behemoth cpu cooler is only designed to cool current oced cpus. so just what in the hell do you think it will take to cool on oced cpu that is already 220 watts out of the box?

How much do you think SB-E will OC to on air, 6GHz You can make a lot better cooler for a CPU then a GPU, period. Unless you want to take up 4 slots.

 

[AtenRa]

they are not true 8 cores and that is a fact

What is a true core ??

 

[ShintaiDK]

How much do you think SB-E will OC to on air, 6GHz You can make a lot better cooler for a CPU then a GPU, period. Unless you want to take up 4 slots.

Yet its harder to cool a CPU due to hotspots than a GPU.

 

[dastral]

"Sorry, but the 8 and 9 series cpus have 8 true cores"
with that comment alone you just proved you are not the least bit knowledgeable

It's even worse than that, all he uses is two AMD PR Slides to "prove his eight cores".
And just insists on lying, which make him look even more stupid, we all know it's not 8 "True Cores"
(A core being defined as an totally independent execution unit that can function on its own).

8T PD/BD design has 8 true integer cores and 4 FP "cores" that can run 2 treads each

Which makes them better than Intel's "bonus cores" as you get a lot more hardware.

It scale way more than 4X when going from single thread
to 8 threads , how could a 4 cores scale more than 4X..?.

Sorry ABWX but you are falling for it too
Of course they will scale better than 4X since you have more than "4 cores".
You have "4 True Cores" and "4 Extra hardware stuff"

Cinebench is really interesting because it provides a lot of data.
- True cores scale almost perfectly :
PII X4 980 has a 1.0 ratio & i5-3470 with a 0.94 ratio & G2120 with a 0.99 ratio
- Intel makes much faster cores (not a surprise)
- AMD Cores "fake cores" have much better scaling :
4+4 on FX8350 has a 0.78 ratio of scaling vs 4+4 on 3770K has a 0.56 ratio of scaling

While the results are to be used with some care, we clearly see that "Fake Cores" (from AMD & Intel) do not even come close to real cores.

A "Real Octo Core" FX8 should get 1.11x8 so around 8.5 or 9, yet it barely gets to 7.
A "Real Octo Core" i7 should get 1.66x8 so around 13, yet it barely gets to 7.5.

When you compare this scaling to PII X4 and 3470 you see both founders lie to you.

Intel gives you "a bit more half than what you were promised" but these are insanely fast.
AMD gives you "80% of what you were promised" but these are hellishly slow.

Food for thought

 

[wand3r3r]

I find this an interesting attempt by AMD to compete. They've been getting beat pretty badly by intel so hopefully their next attempt will actually involve engineering a better chip.

I don't see the point of all the whining, it's going to use some power sure but there are alternatives. I do like the factory guaranteed 5GHz.

The one unknown is the price. I'm assuming reasonable, but if its expensive then I'll have a problem.

 

[AtenRa]

(A core being defined as an totally independent execution unit that can function on its own).

So, 8 independent Integer units are 8 cores.

 

[jhu]

Per your request I ran POV Ray ver 3.7 RC7 on the 8350 stock and at 4.7 Ghz and the 3770k stock and at my OC of 4.4Ghz. I ran a 1 thread render and a 8 thread render. Here we go with results.
For 1 thread
8350 stock 15 minutes 50seconds
8350 @ 4.7 Ghz 15m 47s
3770k stock 15m12s
3770k @ 4.4Ghz 12m 29s

For 8 threads
8350 stock 2m30s
8350 @4.7Ghz 2m31s
3770k stock 3m17s
3770k @4.4Ghz 2m 43s

Hope this was the info you were looking for.

Thanks. Are the stock results with turbo enabled or no?

 

[Piotrsama]

Per your request I ran POV Ray ver 3.7 RC7 on the 8350 stock and at 4.7 Ghz and the 3770k stock and at my OC of 4.4Ghz. I ran a 1 thread render and a 8 thread render. Here we go with results.

For 8 threads
8350 stock 2m30s
8350 @4.7Ghz 2m31s
3770k stock 3m17s
3770k @4.4Ghz 2m 43s

Hope this was the info you were looking for.

My 8350 stock (mem. @ 1866) takes 2m57s.
Sounds as if both times you had it OCed.

How are you OCing it? Through BIOS or some program on Windows?

 

[mrmt]

So we have: much more dedicated die area for GPU, bigger modules that after the 32nm->28nm probably ended the same size (or even smaller), much improved power management with aggressive powering down of caches/cores. I expect Kaveri to have 15-20% higher ST IPC, clock about the same or close to Richland and scale 10% better with more threads (equaling ~30% gain in MT workloads).

That's too much for a typical AMD redesign and a half-node shrink without SOI. I doubt they won't take a hit on frequency because of it.

 

[galego]

Highlight the 47 fps that the 8350 is getting please.

No. I will not mention the same review by fifth time.

 

[guskline]

Thanks. Are the stock results with turbo enabled or no?

With turbo enabled. Do you want me to disable it?

 

[SlowSpyder]

yes you do because that gigantic behemoth cpu cooler is only designed to cool current oced cpus. so just what in the hell do you think it will take to cool on oced cpu that is already 220 watts out of the box?

A CPU cooler also doesn't have to cool VRM's and memory.


Earlier in this thread I ran the Intel burn test with Cool & Quiet turned off and a Kill a Watt. My power use went up some 260 watts on my Phenom II. My four year old CPU cooler with a single 120mm fan and a year and a half old application of AS5 seemed to keep things in check. So hopefully today's coolers are a bit better, but if you really want any headroom you may need water for something like this.

 

[guskline]

I find this an interesting attempt by AMD to compete. They've been getting beat pretty badly by intel so hopefully their next attempt will actually involve engineering a better chip.

I don't see the point of all the whining, it's going to use some power sure but there are alternatives. I do like the factory guaranteed 5GHz.

The one unknown is the price. I'm assuming reasonable, but if its expensive then I'll have a problem.

Agreed. Price is the big issue. I really doubt $200.00 but if it's $350.00 or less it will sell. $800.00 ? No.

 

[galego]

If you are making predictions on the basis of that graph, you cant predict cpu scaling from a gpu limited benchmark in any case. Granted there is a small difference among the cpus, but the test is also running into a gpu bottleneck, so any prediction of cpu scaling would be a wild guess at best.

But that graph gives the same ratio between the extreme chip and the 8350 than another testing specifically CPUs. Therefore there is not gpu bottleneck.

And still my prediction was conservative (I used a 15% scaling). My prediction could vary one or maybe two FPS (I wrote "about") but the FX-9590 will continue being faster.

It's even worse than that, all he uses is two AMD PR Slides to "prove his eight cores".
And just insists on lying, which make him look even more stupid, we all know it's not 8 "True Cores"

I am going to ignore your misleading posts, but not your ad hominem. Apart from the slides I gave a link to Anandtech news about the Centurion, where they also note that the series 8000 and 9000 have "8 cores". The link that I gave in #442 is this

http://www.anandtech.com/show/7066/amd-announces-fx9590-and-fx9370-return-of-the-ghz-race

I gave only the link, but now I will quote a bit:

FX-4000 (two modules/four cores), FX-6000 (three modules/six cores), and FX-8000 (four modules/eight cores) made sense, but FX-9000 breaks that pattern.

The FX-9000 series is now AMD’s highest performance four module/eight core processor for their AM3+ platform

I will ignore further noise about this.